論文

© 2020 IEEE. The large spectral bandwidth at millimeter-wave (mmWave) frequencies provides a mean to achieve very high data rates in wireless communication systems. A unique characteristic of mmWave is that mmWave links are very sensitive to blockage and have large propagation path loss, which exhibits low line-of-sight (LoS) probability, unstable connectivity and unreliable communication. To overcome such challenges, one of the existing solution is to associate the user equipment (UE) with other available Base Stations (BSs) by handover (HO) if the serving BS is blocked. In this paper, for a pedestrian scenario, we propose two reinforcement learning (RL) based user association algorithms, which accounts for the past experience of the blockage on the position of the UE. One focuses on the reward to increase the sum LoS probability and is named as Blockage-Aware User Association (BAUA). The other focuses on the reward to balance the tradeoff between the throughput and the LoS probability and is named as modified BAUA. Simulation results show that the BAUA algorithm increased sum LoS probability and the modified BAUA algorithm show better trade-off between the throughput and the LoS probability than the maximum Signal-to-Interference-plus-Noise Ratio (SINR) based and maximum-throughput based user association algorithms.

© 2018 Elsevier B.V. This study characterizes the overall bit-error-rate (BER) performance optimization of avalanche photodiode (APD) receiver-based amplify-and-forward (AF) relay-added free-space optical (FSO) communication system. The system of interest uses subcarrier intensity quadrature amplitude modulation (QAM) signals and subjects to scintillation due to optical turbulence. A weak atmospheric turbulence (clear-air condition) scenario is considered, for which the received signal intensity fluctuation can be modeled as a log-normal random process. A closed-form expression of the overall BER is derived by taking into account the impact of atmospheric loss, turbulence strength, thermal-noise and shot-noise limited conditions. It can be inferred from the numerical results that using AF relay nodes can extend the transmission distance and minimize BER performance of FSO system significantly compared with the direct transmission. Moreover, the selection of APD's gain values is indispensable to minimize the system's BER performance. Hence, the designed system could achieve the minimal BER by selecting an optimal APD gain value appropriately. In addition, the optimal gain values of APD significantly depend on various conditions, such as noise temperature, link distance, and the number of relay nodes.

© 2018 IEEE. With the rapid growth of social networks and microblogging websites, communication between people from different cultural and psychological backgrounds has become more direct, resulting in more and more 'cyber' conflicts between these people. Consequently, hate speech is used more and more, to the point where it has become a serious problem invading these open spaces. Hate speech refers to the use of aggressive, violent or offensive language, targeting a specific group of people sharing a common property, whether this property is their gender (i.e., sexism), their ethnic group or race (i.e., racism) or their believes and religion. While most of the online social networks and microblogging websites forbid the use of hate speech, the size of these networks and websites makes it almost impossible to control all of their content. Therefore, arises the necessity to detect such speech automatically and filter any content that presents hateful language or language inciting to hatred. In this paper, we propose an approach to detect hate expressions on Twitter. Our approach is based on unigrams and patterns that are automatically collected from the training set. These patterns and unigrams are later used, among others, as features to train a machine learning algorithm. Our experiments on a test set composed of 2010 tweets show that our approach reaches an accuracy equal to 87.4% on detecting whether a tweet is offensive or not (binary classification), and an accuracy equal to 78.4% on detecting whether a tweet is hateful, offensive, or clean (ternary classification).

© 2019 IEEE. Heart rate (HR) variability indicates health condition and mental stress. The development of non-contact HR monitoring techniques with Doppler radar is attracting great attention. However, the performance of heartbeat detection via radar signals easily degrades due to respiration and body motion. In this paper, first, a stochastic gradient approach is applied to reconstruct the high-resolution spectrum of heartbeat by proposing the zero-attracting sign least-mean-square (ZA-SLMS) algorithm. To correct the quantized gradient of cost function and penalize the sparse constraint on updating the spectrum, a more accurate heartbeat spectrum is reconstructed. Then, to better adapt to the noises of different strengths caused by subjects' movements, an adaptive regularization parameter is introduced in the ZA-SLMS algorithm as an improved variant, which can adaptively regulate the proportion between gradient correction and sparse penalty. Moreover, in view of the stability of the location of the spectral peak associated with the HR when the size of time window slightly changes, a time-window-variation (TWV) technique is further incorporated in the improved ZA-SLMS (IZA-SLMS) algorithm for more stable HR estimation. Through the experiments on five subjects, our proposal is demonstrated to bring a significant improvement in accuracy compared with existing detection methods. Specifically, the IZA-SLMS algorithm with TWV achieves the smallest average error of 3.79 beats per minute when subjects type on a laptop.

© 2017 IEEE. The growing environmental awareness coupled with the rising costs of energy have sparked a keen interest in the deployment of energy-efficient communication technologies over the infrastructure of cellular networks. Base stations (BSs) are responsible for the largest portion of power consumption and energy usage in cellular networks. Thus, sleep/wake-up scheduling strategies for BSs can significantly improve the energy-efficiency (EE) of cellular networks. In this paper, we propose a Fuzzy Q-Learning based energy-efficient sleep/wake-up mechanism for BSs in a heterogeneous network. The goal is to save energy, without compromising the offered Quality of Service (QoS), by switching off the redundant BSs according to the local traffic profile and depending on the required area coverage and cell EE. The energy savings achieved under our proposed sleep scheduling strategy may lead to inevitable coverage holes as fewer BSs are active. To this end, we propose to enhance the sleep/wake-up mechanism with device-to-device communications to compensate for the coverage loss in areas where BSs are switched off. Simulation results validate that the proposed framework provides significant improvements in power consumption and the EE while satisfying the minimum coverage and QoS requirements.

© 2018 APSIPA organization. Nonlinear blind source separation is the process of estimating either the original signals or mixture functions from the degraded signals, without any prior information about original sources. The key idea is to recover the sources by estimating an approximation function so as to approximate the inverse of mixing function. However, in practice, the approximation function is derived from some estimation algorithm with finite sample size, which leads to the performance loss. In this paper, we work on the convergence and asymptotic analysis of the separation approach, which uses the flexible approximation to extract the nonlinearity of mixture function so that to make the problem linearly separable. The analysis stems from the performance of a mismatched estimator that accesses the finite sample size. By providing a closed-form expression of normalized mean squared error (NMSE), we can present a novel algebraic formalization that leads to the upper bound on the estimation error. The simulation results show that if the flexible approximation can extract the nonlinearity of mixing functions, the minimized NMSE can be achieved as the sample size tends to be infinity. This implies that the algorithm is feasible to separate the distortion of the nonlinear mixture.

© 2018 IEEE. Interference alignment (IA) is a technique that can suppress interference with a small number of antennas by aligning interference signals using transmit weights. These weights are designed based on the channel state information (CSI) fed back from each receiver, however, under the timevarying channel, the estimated CSI can be delayed/outdated, which will result in an imperfect IA. Therefore, IA with channel prediction has attracted much attention. The auto regressive (AR) model is known as a prediction method that predicts a future state based on only the past states. In the conventional channel prediction based IA method, the past channels are used directly for prediction. Therefore, the number of calculations for prediction can be too large. In this paper, based on the AR model, we describe a low complexity and high accuracy channel prediction method for IA. To predict the future channel, we only use the differences of channels between adjacent times instead of using the past channels directly. This will lead to a very low channel prediction error. Simulations show that the proposed method improves prediction accuracy and requires less calculation than the conventional one. Moreover, the IA with the proposed channel prediction method will achieve a higher transmission rate.

© 2013 IEEE. Bitcoin is one of the most popular decentralized cryptocurrencies to date. However, it has been widely reported that it can be used for investment scams, which are referred to as high yield investment programs (HYIP). Although from the security forensic point of view it is very important to identify the HYIP operators' Bitcoin addresses, so far in the open technical literature no systematic method which reliably collects and identifies such Bitcoin addresses has been proposed. In this paper, a novel methodology is introduced, which efficiently collects a large number of the HYIP operators' Bitcoin addresses and identifies them based upon a novel analysis of their transactions history. In particular, a scraping-based method is first proposed which is able to collect more than 2,000 HYIP operators' Bitcoin addresses from the Internet thus providing a large number of the HYIPs' samples. Second, a supervised machine learning technique, which classifies, whether or not, specific Bitcoin addresses belong to the HYIP operators, is introduced and its performance is evaluated. The proposed classification method is based upon two novel approaches, namely the rate conversion technique that mitigates the effect of Bitcoin price volatility and the sampling technique that reduces the computational amount without sacrificing the classification performance. By employing close to 30,000 real Bitcoin addresses, extensive performance evaluation results obtained by means of computer simulation experiments have shown that the proposed methodology achieves excellent performance, i.e., 95% of the HYIP addresses can be correctly classified, while maintaining a false positive rate less than 4.9%. In order to further validate the proposed classifier's ability to detect the HYIP operators' Bitcoin addresses, our designed classifier has been tested against a recently published list of the HYIP addresses maintaining its excellent detection accuracy by achieving a 93.75% success rate.

© 2020 IEEE. To build a system for monitoring elderly people living alone, an important step needs to be done: identifying the presence/absence of the person being monitored and his location. Such task has several applications that we discuss in this paper, and remains very important. Several techniques were proposed in the literature. However, most of them suffer from issues related to privacy, coverage or convenience. In the current paper, we propose an infrared array sensor-based approach to detect the presence/absence of a person in a room. We used a wide angle low resolution sensor (i.e., 32×24 pixels) to collect heat-related information from the area monitored, and used Deep Learning (DL) to identify the presence of up to 3 people with an accuracy reaching 97%. Our approach also detects of the presence or absence of a person with a 100% accuracy. Nevertheless, it allows identifying the location of the detected people within a room of dimensions 4×7.4 m with a margin of 0.3 m.

© 2017 IEEE. With the huge increases in traffic volumes and subscribers, diverse devices, and rich media applications, manual management of mobile network becomes highly challenging in terms of optimization and management. Self-organizing networks (SON) has been introduced to optimize the network in an automatic manner. In this paper, we address the coverage and capacity joint optimization (CCO) by adaptively and simultaneously adjusting both antenna tilt and power. To this end, we propose: · a multi-player multi-armed bandit (MAB) framework (decentralized restless upper confidence bound (RUCB) algorithm) with a change point detection test based on Page-Hinkley (PH) statistics used to decide whether some change has occurred in the environment. Then, the strategy is designed to deal with such a change. · a central unit to deal with simultaneous conflicting actions when many cells decide to start the optimization process at the same time. · a Pareto search framework to deal with multi-objective optimization (CCO). To evaluate our work, we compared our proposal with the fixed antenna parameter scheme and with the linear scalarization function that transforms the multi-objective optimization problem into a scalar function. Simulation results show that the proposed method could improve user experience in terms of cell-center capacity and cell-edge coverage compared to different conventional methods and under different number of users.

© 1964-2012 IEEE. In non-contact heart rate (HR) monitoring via Doppler radar, the disturbance from respiration and/or body motion is treated as a key problem on the estimation of HR. This paper proposes a blind source separation (BSS) approach to mitigate the noise effect in the received radar signal, and incorporates the sparse spectrum reconstruction to achieve a high-resolution of heartbeat spectrum. The proposed BSS decomposes the spectrogram of mixture signal into original sources, including heartbeat, using non-negative matrix factorization (NMF) algorithms, through learning the complete basis spectra (BS) by a hierarchical clustering. In particular, to exploit the temporal sparsity of heartbeat component, two variants of NMF algorithms with sparseness constraints are applied as well, namely sparse NMF and weighted sparse NMF. Compared with usual BSS, our proposed BSS has three advantages: 1) clustering-induced unsupervised manner; 2) compact demixing architecture; and 3) merely requiring single-channel input data. In addition, the HR estimation method using our proposal delivers more satisfactory precision and robustness over other existing methods, which is demonstrated through the measurements of distinguishing people's activities, gaining both smallest absolute errors of HR estimation for sitting still and typewriting.

© 2019 IEEE. Respiration is known to reflect our health condition, which motivates researchers to develop various radar-based respiration rate estimation methods. However, these conventional methods do not work, when a subject is not right in front of the radar. In this paper, we propose a novel CNN (Convolutional Neural Network)-based respiration rate estimation method in indoor environments via a MIMO (Multiple-Input Multiple-Output) FMCW (Frequency Modulated Continuous Wave) radar. A MIMO FMCW radar can estimate the DoA (Direction of Arrival) and the distance between a MIMO FMCW radar and an object. Thus, the respiration can be captured based on the phase variation at a subject's location. However, even when the advanced signal processing, e.g., MUSIC (MUltiple SIgnal Classification) algorithm, is used, it is difficult to estimate the DoA and the distance in indoor environments due to the large effect of multipath. To deal with this problem, in the proposed method, phase variations against various locations are calculated from the received signals of a MIMO FMCW radar, and then spectrograms are calculated from the phase variations. Each spectrogram is subsequently fed into the CNN that outputs the respiration rates, e.g., 0.1 Hz, 0.2 Hz, and non-respiration, i.e., a spectrogram without the effect of respiration, where is one of the deep learning techniques that have been successfully applied to the image recognition. Through the experiments we confirmed that except for when microwaves were not transmitted directly toward a subject's chest due to furniture, the proposed method accurately estimated the respiration rate, regardless of the situation.

© 2018 IEEE. A physical layer network coding (PLNC) is a highly efficient scheme for exchanging information between two nodes. Since the relay receives the interfered signal between two signals sent by two nodes, it hardly decodes any information from received signal. Therefore, the secure wireless communication link to the untrusted relay is constructed. The two nodes optimize the transmit power control for maximizing the secure capacity but these depend on the channel state information informed by the relay station. Therefore, the untrusted relay disguises the informed CSI for exploiting the information from two nodes. This paper constructs the game of two optimizations between the legitimate two nodes and the untrusted relay for clarifying the security of PLNC with untrusted relay.

© 2018 IEEE. In this paper, due to the unlimited growth of service demands and high load traffic, a green wireless communication system has been considered with the aim of decreasing energy consumption of heterogeneous networks (HetNets). This system provides a proportion of required energy of Base Stations (BSs) by employing some renewable power suppliers (RPSs), while user equipments (UEs) especially benefit from short device-to-device (D2D) links as a promising technique to design energy-efficient HetNets. Indeed, we attempt to focus on efficient interaction between the RPS's random green power generations and the BS's dynamic of energy consumption via minimization of an energy-based cost function, considering the D2D impacts on efficient utilization of energy and bandwidth. Simulation results validate that the proposed framework provides significant improvements in electrical grid power-consumption and total energy efficiency (EE) of the network.

© 2020 IEEE. In a massive multiple-input multiple-output (MIMO) system, belief propagation (BP) detection is known as a method to separate and detect received signals. In BP detection, a MIMO channel is represented by a factor graph and the transmitted symbols are estimated by message passing. However, the convergence property of BP deteriorates due to multiple loops included in the MIMO channel. As a method to improve the convergence property and the detection performance, the damped BP that averages two successive messages with a weighing factor (called damping factor) is known. To train the damping factors off-line for each antenna configuration, deep neural network-based damped BP (DNN-dBP) has been reported. The problem with DNN-dBP is that the detection performance deteriorates when there is a difference in the channel correlation between training and test. This is because the optimal damping factors vary with the channel correlation. In this paper, to solve this issue, we derive the damping factors of BP with the node selection (NS) method that selects nodes to be updated to lower spatial correlation using DNN-dBP. By applying the NS method, the channel correlation among the selected nodes in BP detection is lowered. Therefore, the proposed method can reduce the detection performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. In addition, the convergence property of BP is improved by applying the NS method. Therefore, the proposed method can improve the detection performance compared to the conventional DNN-dBP with the same computational complexity. By computer simulation, it is shown that the proposed method significantly reduces the bit error rate (BER) performance deterioration due to the mismatches of the channel correlations between training and test in DNN-dBP. The results also show that the proposed method has the better BER performance than the conventional DNN-dBP with the same computational complexity.

© 2020 IEEE. So far, most separation approaches of vital signs such as heartbeat and respiration, are implemented based on linear mixtures. However, some literatures have reported that non-linear mixtures actually occur in the associated applications, e.g., heart rate (HR) estimation with Doppler radar, where the simple linear demixing architecture may limit the effect of source separation. In addition, the human motions during HR measurement further complicate the mixing processes. The issue motivates us to exploit a more suitable separation approach to deal with contact-free HR estimation, considering non-linear mixtures including motions. A semi-supervised deep clustering (DC) is proposed to separate the three mixed sources of heartbeat, respiration, and motions, by segmenting the spectrogram of Doppler signal. First, through training a deep recurrent neural network (RNN) with long short-term memory (LSTM) via heartbeat/respiration-only data, the embeddings to each frame-sample from spectrogram can be acquired, which enables feature optimization in a lower dimensional space. Then, in the test phase, K-means clusters the embeddings associated with each source, to infer the masks used for spectrogram segmentation. The proposed deep clustering has three main strengths: It (i) gets rid of the restriction of mixture class, relying on data mining; (ii) can handle three-source mixtures by training two sorts of source-independent samples; (iii) only requires the mixtures from single-channel. The HR measurement experiments on subjects' sitting still and typing, validate the improvements of accuracy and robustness by our proposal, over some prevailing approaches in signal decomposition or separation.

© 2014 IEEE. With the high penetration of large-scale distributed renewable energy generation, the power system is facing enormous challenges in terms of the inherent uncertainty of power generation of renewable energy resources. In this regard, virtual power plants (VPPs) can play a crucial role in integrating a large number of distributed generation units (DGs) more effectively to improve the stability of the power systems. Due to the uncertainty and nonlinear characteristics of DGs, reliable economic dispatch in VPPs requires timely and reliable communication between DGs, and between the generation side and the load side. The online economic dispatch optimizes the cost of VPPs. In this article, we propose a deep reinforcement learning (DRL) algorithm for the optimal online economic dispatch strategy in VPPs. By utilizing DRL, our proposed algorithm reduced the computational complexity while also incorporating large and continuous state space due to the stochastic characteristics of distributed power generation. We further design an edge computing framework to handle the stochastic and large-state space characteristics of VPPs. The DRL-based real-time economic dispatch algorithm is executed online. We utilize real meteorological and load data to analyze and validate the performance of our proposed algorithm. The experimental results show that our proposed DRL-based algorithm can successfully learn the characteristics of DGs and industrial user demands. It can learn to choose actions to minimize the cost of VPPs. Compared with the deterministic policy gradient algorithm and DDPG, our proposed method has lower time complexity.

© 2017 IEEE. Bitcoin is a decentralized currency system that does not need any central authorities. All transactions issued by users have been recorded in the common ledger, called blockchain, which is shared by all users. In Bitcoin, an SPV (Simplified Payment Verification) client, which is a lightweight client that does not possess the entire blockchain, are developed for storage constrained devices such as a mobile phone. For an SPV client to check if there are transactions related to it, a Bloom filter where their Bitcoin addresses are involved is sent to a full client that possesses the entire blockchain. The full client only transfers transactions of which Bitcoin addresses are positive on the received Bloom filter. However, it is necessary to preserve the privacy of SPV clients when designing a Bloom filter because SPV clients' Bitcoin addresses will be identified by a full client with high probability. In this paper, we propose a privacy-preserving Bloom filter design for SPV clients based on γ-Deniability. γ-Deniability is a privacy metric that shows how much true positive Bitcoin addresses are hidden by the false positives in a Bloom filter. Furthermore, in order to design a Bloom Filter that satisfies a certain γ-Deniability, it is necessary to know the number of unique Bitcoin addresses that appear for the first time since the queried time. Based on our manual inspection, we propose to estimate it based on the linear regression. We show that our scheme achieves good estimation accuracy and γ through the simulation with a real Bitcoin blockchain.

© 2020 IEEE. Polar codes are attracting much attention and being used for control channels of the 5th generation of mobile communication system (5G). As a feature, it is easier to implement encoder and decoder than Turbo codes and LDPC (Low Density Parity Check) codes. One of the decoding methods of polar codes is BP (Belief Propagation) decoding, which can decode in parallel, so that decoding can be performed at high speed. However, due to hardware limitation, calculations on the decoder get very complicated. This issue can be solved by using the information bottleneck method. This method compresses an observation variable to a quantized one while attempting to preserve the mutual information shared with a relevant random variable. In the conventional research, the information bottleneck method is applied to the BP decoding of the LDPC codes. In this paper, the information bottleneck method is used for the BP decoding of polar codes. The BP decoding of polar codes is distinct from that of LDPC codes. It has several types of the messages, and each time a message is updated, the decoding becomes more complex. By using the information bottleneck method, the decoder can compress the channel outputs and the messages of BP into unsigned integers while preventing degradation of the error correcting performance. Thus, we can reduce the complexity of calculation in the decoding process and easily implement the decoder. This paper also investigates the minimum bit width for quantization with negligible degradation and the suboptimal E-{b}/N-{0} for designing the lookup tables. These lookup tables are used for updating the messages. The simulation results show that the error correcting capability of the discrete polar decoders of the proposed method is negligibly degraded compared to the BP decoding without compression.

© 2018 IEEE. The number of elderly people who live alone is increasing in many countries. Furthermore, many of their accidents occur at home. Hence, it is an urgent demand for a system monitoring their activities to detect accidents indoor such as falling. In conventional systems of fall detection using array antennas, falling after standing still can be detected with high accuracy by leveraging the features indicating the change of radio wave propagation. However, it is difficult to detect falling after walking correctly. In this paper, to improve fall detection accuracy including falling after walking, we propose an accurate fall detection system by leveraging the features indicating the change of Doppler signals during human activities in detail. Analyzing Doppler signals is useful to detect falling since they are observed when a radio wave reflects by moving objects. We conducted experiments in actual rooms to demonstrate that the proposed method can detect falling after both standing and walking with high accuracy.